GitLab

The new Skia recording backend is not as sensitive to the number of draw
ops as the old one -> we don't have to bend over backwards to suppress
unneeded saves at the GC level.

While this results in a slight increase in the number of recorded ops,
the unnecessary saves and restores are optimized out by the bounding box
hierarchy - so there is no playback/rasterization time penalty (Skia
also has an explicit save/restore optimizer which can be used for this
purpose if we ever decide to drop the BBH).

Local perf measurements show no rasterization impact and a minor (<1%)
improvement for record time.

R=schenney@chromium.org,senorblanco@chromium.org

Review URL: https://codereview.chromium.org/653153004

git-svn-id: svn://svn.chromium.org/blink/trunk@183750 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This is another regression from
https://codereview.chromium.org/570763003 that wasn't
solved by https://codereview.chromium.org/598113003
as it missed more subtleties: most callers want the
clipped selection bounds. Only one caller actually
cared about the unclipped bounds:
FrameSelection::revealSelection as it needs to know
where the outside-the-viewport-selection is.

This change splits bounds() in 2, depending on
the intended behavior. While touching the code,
switched it to use LayoutRect as none of the
callers cares about FloatRect.

This change fixes
fast/transforms/selection-bounds-in-transformed-view.html
that got broken by using a fractional scroll offset.

Unfortunately no extra test as the issue requires
generating a gigantic drag-image that
DRT doesn't do.

BUG=415917

Review URL: https://codereview.chromium.org/640023002

git-svn-id: svn://svn.chromium.org/blink/trunk@183749 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=423320
R=pfeldman, loislo

Review URL: https://codereview.chromium.org/660483002

git-svn-id: svn://svn.chromium.org/blink/trunk@183748 bbb929c8-8fbe-4397-9dbb-9b2b20218538

The patch kills invalid test script-window-close-breakpoint.html.
The test itself expoits an old bug crbug.com/6773 and closes
main window instead of a programmatically created one. This
causes the layout test driver to crash on the following
test as there's no window to navigate.

BUG=423364

Review URL: https://codereview.chromium.org/656083002

git-svn-id: svn://svn.chromium.org/blink/trunk@183747 bbb929c8-8fbe-4397-9dbb-9b2b20218538

The additional transform computed by the helper function
RenderSVGResourceContainer::computeResourceSpaceTransform by each
paint-server (that is CTM-dependent), only depends on the input to
preparePaintServer, and not the paint-server itself.
Hoist the computation into SVGRenderSupport::updateGraphicsContext.
This in turn mean that the various preparePaintServer implementations
no longer use the passed RenderStyle (previously only used for checking
for vector-effect) and RenderSVGResourceModeFlags, so remove those.
computeResourceSpaceTransform w/ associated helpers are moved to
SVGRenderSupport.

BUG=420022

Review URL: https://codereview.chromium.org/652243002

git-svn-id: svn://svn.chromium.org/blink/trunk@183746 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Revert of [DevTools] Inject element into toolbox window, do not load modules there. (patchset #2 id:150001 of https://codereview.chromium.org/653613002/)

Reason for revert:
Breaks all references to "document".

Original issue's description:
> [DevTools] Inject element into toolbox window, do not load modules there.
> 
> BUG=none
> 
> Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=183669

TBR=pfeldman@chromium.org
NOTREECHECKS=true
NOTRY=true
BUG=none

Review URL: https://codereview.chromium.org/660503002

git-svn-id: svn://svn.chromium.org/blink/trunk@183744 bbb929c8-8fbe-4397-9dbb-9b2b20218538

When a RenderObject with a non-default touch action is about to be
destroyed, we remove the fake event handler that implements the touch
action from the event handler registry. However the handler may have
already been removed previously by the host Document (e.g., from
Document::open()), so this code can cause the same handler to get
removed twice.

This patch fixes the issue by only removing the handler conditionally
if it is still there.

BUG=419420

Review URL: https://codereview.chromium.org/647503002

git-svn-id: svn://svn.chromium.org/blink/trunk@183743 bbb929c8-8fbe-4397-9dbb-9b2b20218538

It was broken in https://chromium.googlesource.com/chromium/blink/+/30f11148be4ef1037f311e86a7707396e457d429%5E%21/#F13

BUG=89845,403007

Review URL: https://codereview.chromium.org/655203002

git-svn-id: svn://svn.chromium.org/blink/trunk@183742 bbb929c8-8fbe-4397-9dbb-9b2b20218538

TBR=haraken,oilpan-reviews
BUG=421948
NOTRY=true

Review URL: https://codereview.chromium.org/655213002

git-svn-id: svn://svn.chromium.org/blink/trunk@183741 bbb929c8-8fbe-4397-9dbb-9b2b20218538

TBR=haraken,oilpan-reviews
BUG=397902
NOTRY=true

Review URL: https://codereview.chromium.org/656093002

git-svn-id: svn://svn.chromium.org/blink/trunk@183740 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This is because injected stylesheet is removed by obsolete code.

BUG=

Review URL: https://codereview.chromium.org/660443002

git-svn-id: svn://svn.chromium.org/blink/trunk@183739 bbb929c8-8fbe-4397-9dbb-9b2b20218538

R=machenbach@chromium.org,ojan@chromium.org
BUG=416182

Review URL: https://codereview.chromium.org/657533002

git-svn-id: svn://svn.chromium.org/blink/trunk@183738 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This is to replace the existing |SVGElement::parseAttribute| implementation,
but as there are many callers, we would like to proceed this incrementally.
a) Introduce |SVGElement::parseAttributeNew| *LANDED*
b) Convert all overrides |SVGElement::parseAttribute| to
   a |parseAttributeNew| call. <-- This patch
c) Replace existing |parseAttribute| with |parseAttributeNew|, and remove all
   overrides simply calling |parseAttributeNew|

This patch is (b).

BUG=397902

Review URL: https://codereview.chromium.org/643703006

git-svn-id: svn://svn.chromium.org/blink/trunk@183736 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Console Message Storage was moved to frame host because remote frame which used in site-per-process mode not contains Frame Console.
Instrumentation Code generator supports agent dispatching by different objects e.g. messagesCleared dispatch by execution context for workers and by FrameHost for pages.

R=vsevik@chromium.org
BUG=421948

Review URL: https://codereview.chromium.org/645693003

git-svn-id: svn://svn.chromium.org/blink/trunk@183735 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=417275

Review URL: https://codereview.chromium.org/660463002

git-svn-id: svn://svn.chromium.org/blink/trunk@183734 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Update the UI to display "Paused on promise rejection" status instead
of "Paused on exception" when appropriate.

R=pfeldman@chromium.org, yurys@chromium.org

Review URL: https://codereview.chromium.org/651533003

git-svn-id: svn://svn.chromium.org/blink/trunk@183733 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Only used in SVGTextRunRenderingContext::drawSVGGlyphs. Fold up and
simplify. Clean up some includes.

BUG=420022

Review URL: https://codereview.chromium.org/657553002

git-svn-id: svn://svn.chromium.org/blink/trunk@183732 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=420492
R=sigbjornf@opera.com

Review URL: https://codereview.chromium.org/656653003

git-svn-id: svn://svn.chromium.org/blink/trunk@183731 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Make sure DatabaseThread::m_openDatabaseSet has no backing store
when a database thread is terminated.
m_openDatabaseSet outlives a database thread, and its backing store
is allocated in the database thread.  So, we should unlink the
backing store before the thread termination.

* DatabaseThread::cleanupDatabaseThread
 It's possible m_openDatabaseSet.size() is zero and
 m_openDatabaseSet still has backing store in the following scenario:
  1. A web page opens a database.  m_openDatabaseSet has it.
  2. Chromium calls DatabaseTracker::closeDatabasesImmediately()
  3. It calls m_openDatabaseSet.remove().

* DatabaseThread::recordDatabaseOpen
  Don't add a Database object after termination request.
  I think this can't happen. But we change it just in case.

This CL has no tests. It's very hard to make an automated test for the
scenario.

BUG=423271

Review URL: https://codereview.chromium.org/657023003

git-svn-id: svn://svn.chromium.org/blink/trunk@183730 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=360466
TEST=added form-pseudo-valid-style.html layout test

Review URL: https://codereview.chromium.org/646543004

git-svn-id: svn://svn.chromium.org/blink/trunk@183729 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This CL removes member variables of PumpSession, as they are no longer used.
Since r183572, all asynchronous parsing uses background thread and these
state variables were only used for deprecated asynchronous main thread parsing.

BUG=421189

Review URL: https://codereview.chromium.org/657753002

git-svn-id: svn://svn.chromium.org/blink/trunk@183728 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=388375
TEST=n/a

Review URL: https://codereview.chromium.org/653813002

git-svn-id: svn://svn.chromium.org/blink/trunk@183727 bbb929c8-8fbe-4397-9dbb-9b2b20218538

http://src.chromium.org/viewvc/blink?view=revision&revision=183720

BUG=409503
TBR=tkent@chromium.org

Review URL: https://codereview.chromium.org/653333003

git-svn-id: svn://svn.chromium.org/blink/trunk@183726 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Since r183572, SynchronousMode is alywas ForceSynchronous, as all
asynchronous parsing now uses background thread.

This CL removes the enum and removes dead codes.

BUG=421289

Review URL: https://codereview.chromium.org/659713002

git-svn-id: svn://svn.chromium.org/blink/trunk@183725 bbb929c8-8fbe-4397-9dbb-9b2b20218538

NOTRY=true
BUG=417275

Review URL: https://codereview.chromium.org/596323002

git-svn-id: svn://svn.chromium.org/blink/trunk@183724 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=413249
TBR=loislo@chromium.org
NOTRY=true

Review URL: https://codereview.chromium.org/658683002

git-svn-id: svn://svn.chromium.org/blink/trunk@183723 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=412737

Review URL: https://codereview.chromium.org/654923002

git-svn-id: svn://svn.chromium.org/blink/trunk@183722 bbb929c8-8fbe-4397-9dbb-9b2b20218538

The NodeList hierarchy is accessed only by the main thread.
Given that NodeList is performance-sensitive in various benchmarks in blink_perf,
this CL explicitly marks the NodeList hierarchy MainThreadOnly.

BUG=420515

Review URL: https://codereview.chromium.org/646833003

git-svn-id: svn://svn.chromium.org/blink/trunk@183721 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Update some tests in fast/forms/suggestion-picker/ so that they use
openPicker() instead of openPickerWithMock().

BUG=409503

Review URL: https://codereview.chromium.org/633823003

git-svn-id: svn://svn.chromium.org/blink/trunk@183720 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=413249
TBR=loislo@chromium.org
NOTRY=true

Review URL: https://codereview.chromium.org/650783004

git-svn-id: svn://svn.chromium.org/blink/trunk@183719 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Service Workers' Cache API lets authors to fully manage their content caches for offline use.
https://slightlyoff.github.io/ServiceWorker/spec/service_worker/#cache-objects
But Service Worker is planned to be shipped without this API for the first launch.
So we add its own runtime enabled feature.

BUG=423601

Review URL: https://codereview.chromium.org/657723002

git-svn-id: svn://svn.chromium.org/blink/trunk@183718 bbb929c8-8fbe-4397-9dbb-9b2b20218538

A [TypeChecking=interface] attribute has been checked already in its setter. 
If it is not nullable, just call toImpl instead of toImplWithTypeCheck to avoid 
redundant type check.

This change can improve performance and save power.

BUG=

Review URL: https://codereview.chromium.org/654473002

git-svn-id: svn://svn.chromium.org/blink/trunk@183717 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Fix bug in line layout where it was incorrectly assumed that trailing
spaces would always be measured with the simple font code path.

R=chrishtr@chromium.org
BUG=423274
TEST=fast/text/international/complex-text-trailing-space.html

Review URL: https://codereview.chromium.org/655073002

git-svn-id: svn://svn.chromium.org/blink/trunk@183716 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Since r183572, the flag is always true.
This CL removes the flag and dead codes referencing the flag.

BUG=421289

Review URL: https://codereview.chromium.org/657453003

git-svn-id: svn://svn.chromium.org/blink/trunk@183715 bbb929c8-8fbe-4397-9dbb-9b2b20218538

http://src.chromium.org/viewvc/blink?view=revision&revision=183710

BUG=409503
TBR=tkent@chromium.org

Review URL: https://codereview.chromium.org/654083004

git-svn-id: svn://svn.chromium.org/blink/trunk@183714 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Together with the Chromium-side patch, makes top controls work properly
under the virtual viewport mode of pinch zoom. The key changes on the
Blink side are to adjust both the PinchViewport and FrameView max scroll
bounds to account for the change in the top controls offset on the
compositor. Additionally, I changed these to make the same floating point
and snapping calculations as in the compositor to prevent losing pixel
fractions.

Chromium side: https://codereview.chromium.org/641873003/

Note, this was previously landed in 183639 and reverted in 183653 due
to failing tests on Mac bots.

BUG=364109

Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=183639

Review URL: https://codereview.chromium.org/643473002

git-svn-id: svn://svn.chromium.org/blink/trunk@183713 bbb929c8-8fbe-4397-9dbb-9b2b20218538

On Android, a premultiplied ImageBuffer source has RGBA pixels. Import
its bitmap for encoding in that form, rather than BGRA.

Covered by fast/canvas/canvas-toDataURL-webp.html

BUG=421454

Review URL: https://codereview.chromium.org/645333002

git-svn-id: svn://svn.chromium.org/blink/trunk@183712 bbb929c8-8fbe-4397-9dbb-9b2b20218538

My patch https://src.chromium.org/viewvc/blink?view=rev&revision=183623
wrongly assumed that the fullscreen element was on the main frame.  Use
the frame tracked by FullscreenController instead.

NOTRY=true
BUG=423457

Review URL: https://codereview.chromium.org/647343003

git-svn-id: svn://svn.chromium.org/blink/trunk@183711 bbb929c8-8fbe-4397-9dbb-9b2b20218538

If we switch openPicker() in date-suggestion-picker-key-operations.html, it
crashes in PageAnimator::serviceScriptedAnimations because Page is deleted in an
event handler. We need to protect the PageAnimator.

BUG=409503,421321

Review URL: https://codereview.chromium.org/637243002

git-svn-id: svn://svn.chromium.org/blink/trunk@183710 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=412088

Review URL: https://codereview.chromium.org/648423002

git-svn-id: svn://svn.chromium.org/blink/trunk@183709 bbb929c8-8fbe-4397-9dbb-9b2b20218538